Vital signs of a subject or a patient and in particular the respiration rate of a subject can be monitored remotely using a contactless sensor such as a video camera. A general method for determining a respiration rate from image data by means of pattern detection is known from WO 2012/140531 A1. Since the subject of interest can be located freely in the field of view of the camera, the subject and the relevant area from which the vital signs should be derived have to be detected and defined for extraction of the desired vital sign information. Further, different persons in the field of view have to be identified and distinguished for a precise remote measurement.
The traditional identification of the region of interest in general is based on the detection of the human being, e.g. the face or the chest or by using background segmentation.
For the identification of a human being and for measuring vital signs such as a pulse or a respiration rate from a region of interest, US 2009/01411241 suggests to detect the contour segmenting of an infrared video segment to select the region of interest representing a portion of a subject to be measured.
Further, US 2009/0226043 A1 discloses a method for detecting respiratory patterns, wherein the region of interest and in particular the person of interest or different persons of interest are identified on the basis of respiratory pattern changes which indicate behavioral deviations of the persons to be detected.
The disadvantage of the known identification methods is that the camera has to be fixed and additional background information is necessary in order to detect the human beings in the field of view. Further, the known methods are not applicable or not reliable if the persons in the field of view are partially occluded or if the person is obstructing the background before the background has been identified by the camera.